Performance Analysis of Genetic Algorithm and Particle Swarm Applied Individually to the Portfolio Optimization Problem

The Portfolio Optimization Problem consists of determining the best allocation of resources in financial assets, aiming to maximize profits and minimize risks. Nowadays, in the context of Computational Optimization, it was observed that its classic approach, which used a simplistic view of market, gained restrictions that brought it closer to reality, such as, for example, constraints regarding cardinality, lower and upper bounds for each asset. However, the addition of these restrictions increased the complexity of the problem, therefore, the use of metaheuristic procedures became more appropriate given their characteristics. In this context, the present work sought to compare the performance of the Particle Swarm (PSO) and Genetic Algorithm (GA) procedures when applied to the Portfolio Optimization Problem. To carry out the experiments, classic instances from the literature were used and, after 75 cases of analysis, it is found that the Genetic Algorithm demonstrated better performance.